train_llava保存processor的时候出现错误

[1ittlesnow]

qwen_model_name_or_path = "/data/workspace/huggingface/Qwen2-1.5B-Instruct"
autoprocessor = AutoProcessor.from_pretrained(clip_model_name_or_path)
autoprocessor.save_pretrained("huggingface/model003")

通过这个保存了processor之后

autoprocessor1 = AutoProcessor.from_pretrained("huggingface/model001")
只能通过AutoProcessor加载，不能通过llavaProcessor加载